Struct OllamaProvider 

pub struct OllamaProvider {
    pub base_url: String,
    pub model: String,
}

Local chat provider for OpenAI-compatible servers (Ollama, LM Studio, llama.cpp’s server, vLLM, etc.).

Why: developers increasingly run a local model server during dev to avoid API costs and latency. The OpenAI-compatible /v1/chat/completions endpoint with SSE streaming is the de-facto common denominator. What: stores the server’s base URL and the model id to request. chat_stream POSTs {model, messages, tools?, stream: true} and parses SSE data: frames identically to the OpenRouter path. Test: shape covered by ollama_provider_reports_metadata; streaming and tool-call accumulation by ollama_provider_streams_sse_deltas and accumulates_streamed_tool_call_fragments.

Fields

base_url: String

model: String

Implementations

impl OllamaProvider

pub fn new(base_url: impl Into<String>, model: impl Into<String>) -> Self

Construct a provider from a base URL and model id.

Why: parallel to OpenRouterProvider::new so callers see a consistent shape across providers. What: stores both fields verbatim; the base URL should NOT have a trailing slash — the implementation appends /v1/chat/completions. Test: covered by ollama_provider_reports_metadata.

Trait Implementations

impl ChatProvider for OllamaProvider

fn name(&self) -> &str

Human-readable provider name (e.g. "openrouter", "ollama").

fn model(&self) -> &str

Model identifier sent on every request.

fn chat_stream<'life0, 'async_trait>( &'life0 self, messages: Vec<ChatMessage>, tools: Vec<ToolDef>, tx: Sender<ChatEvent>, ) -> Pin<Box<dyn Future<Output = Result<()>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Stream chat events into tx. tools empty disables tool use.